Commits

Branch selector

 master

User selector

All users

Datepicker

All time

Commit History

Commits on Sep 8, 2021

• Create README.md

  little-nem

  authored
  8cf4f0c

  Copy full SHA for 8cf4f0c

Commits on Jun 7, 2019

• Less frequents tests

  little-nem

  committed
  6c69b6c

  Copy full SHA for 6c69b6c

• Fixing the friction parameter

  little-nem

  committed
  b4c82ce

  Copy full SHA for b4c82ce

• Log the critic gradients (check if there is gradient explosion during training)

  little-nem

  committed
  7465787

  Copy full SHA for 7465787

• Using the fetch slide environment

  little-nem

  committed
  f6638b7

  Copy full SHA for f6638b7

• Add modified fetch_slide env

  little-nem

  committed
  65a4245

  Copy full SHA for 65a4245

Commits on Jun 6, 2019

• Smaller replay buffer

  little-nem

  committed
  fa3faa6

  Copy full SHA for fa3faa6

• Add critic loss logging during training

  little-nem

  committed
  ba39835

  Copy full SHA for ba39835

Commits on Jun 5, 2019

• Keep an unlimited amount of checkpoints during training (old would only keep 5 of them...)

  little-nem

  committed
  54cac4b

  Copy full SHA for 54cac4b

• Add plain text (csv) logging of the success_rate (more readable outside of tensorboard)

  little-nem

  committed
  b6d63e2

  Copy full SHA for b6d63e2

• Train with dense reward

  little-nem

  committed
  a72fe13

  Copy full SHA for a72fe13

• Removing a useless saver

  little-nem

  committed
  16891e2

  Copy full SHA for 16891e2

• Refactor model loading and saving (getting rid of the tflearn.dnn wrapper...)

  little-nem

  committed
  e3cea00

  Copy full SHA for e3cea00

• Add model loading and unrolling

  little-nem

  committed
  d27b43e

  Copy full SHA for d27b43e

• Initialization that makes more sense

  little-nem

  committed
  47c0a0a

  Copy full SHA for 47c0a0a

• Add success_rate logging and saving model checkpoints during training

  little-nem

  committed
  8f547da

  Copy full SHA for 8f547da

Commits on Jun 4, 2019

• Fixing a typo

  little-nem

  committed
  2fc60b9

  Copy full SHA for 2fc60b9

• More training time

  little-nem

  committed
  b287fcb

  Copy full SHA for b287fcb

• Added Hindsight Experience Replay, to deal with sparse reward

  little-nem

  committed
  b84190c

  Copy full SHA for b84190c

• Add noise on the actions during training

  little-nem

  committed
  c389a49

  Copy full SHA for c389a49

• Added testing during training to evaluate the current training status

  little-nem

  committed
  5d96047

  Copy full SHA for 5d96047

• Refactoring, and now use batch of episodes instead of batch of transitions for training

  little-nem

  committed
  42af539

  Copy full SHA for 42af539

Commits on May 31, 2019

• Initial commit

  little-nem

  committed
  63c7972

  Copy full SHA for 63c7972